New Zealand Institute for the study of competition and regulation inc.

Abstract

The Ultimatum game is simple and this facilitates its use in the study of predictions of game theory. Experimental evidence suggests that it does
not predict individual behavior well, unless individuals gain welfare
from fairness in transactions, or have expectations about some wider
game. Our model excludes any notion of fairness by including (potential)
rivalry in transactions. In this game the proposer's expectations yield
outcomes that are consistent with experimental evidence. Offers can be
large or small, with none in an intermediate range. The consequent distribution appears in Dictator game experiments. Our model explains how it is generated by expectations.